Real-Time Human Motion Detection, Tracking and Activity Recognition with Skeletal Model

Human activity recognition with 3Dskeletal model has been attracted in a lot of application area. Representations of human based on 3D perception have been occurred prevalent problems in activity recognition. In recent work with RGB-Depth cameras, expensive wearable sensors and illuminator array have been used to construct the 3D human skeleton model in recognition system. But these systems have been defined specific lightening condition, limited range. and great constraint in outdoor applications. To overcome this restriction, the proposed system is considered on the real-time video sequences of the human movement to understand human behavior in indoor and outdoor environment. The proposed method is constructed human detection and motion tracking by using framewise displacement and recognition is based on skeletal model with deep learning framework. The result is to become an efficient detection, tracking and recognition system for real-time human motion. The performance and accuracy of the system is analyzed with the various videos to show the results.